Impact Localization of Composite Structure Based on SSA-CNN-LSTM Using FBG Sensors

Composite material structures are widely used in aerospace, aviation, and other industries. These structures are always in service as the main component of the structure and are susceptible to damage from impact loads throughout their manufacturing, transportation, and service processes. Structural health monitoring of composite material structures is, therefore, crucial for preventing structural failure and potential catastrophic consequences. In this article, an impact localization system for composite material structure based on fiber Bragg grating (FBG) sensors is established. The wavelet packet decomposition method is used to analyze the time-frequency domain characteristics of the impact signal and principal component analysis (PCA) is used to diminish the dimensionality of the wavelet packet energy spectrum, enabling the extraction of features that effectively characterize the location of impact points. The sparrow search algorithm (SSA)-convolutional neural network (CNN)-long short-term memory (LSTM) neural network model is employed for impact localization. The performance of employing PCA for feature selection and impact localization is assessed and compared across various wavelet bases and decomposition layers. Finally, the proposed impact localization method based on SSA-CNN-LSTM is validated on a composite tank structure with an experimental area of $400\times 400$ mm, and the average localization error under the nine-layer wavelet packet decomposition is 10.1 mm. The experimental results demonstrate that this method requires a low FBG demodulation speed and offers superior localization accuracy for impacts on composite structures compared to traditional models.